[unreadable] Project Summary: 518 human protein kinases modulate the activity of -30% of all proteins, and collectively control almost all complex pathways and decisions of a cell. Despite tremendous experimental analysis on some kinases, we know little of the detailed function of most members of this uniquely important family. We propose to use genome sequences to tap hundreds of millions of years of evolutionary experimentation, in order to clarify the link between kinase sequence and biological function. We will use our extensive knowledge from the discovery of the human and mouse kinomes to predict all kinase orthologs in up to 40 vertebrate genomes. We will then map the evolutionary constraints on every residue of every kinase, and predict domains, motifs, phosphorylation sites, and other functional regions of proteins, extending the kinome catalog to unprecedented resolution and generating a wealth of hypotheses for experimental testing. Finally, we will apply this knowledge to predict the functional impact of both SNPs and somatic mutations in cancer, providing a valuable preview of the proposed cancer genome atlas. [unreadable] [unreadable] By using kinases as a model family to explore the predictive power of comparative genomics, we will develop well-validated tools and parameters which can then be applied to any human gene or gene family. [unreadable] [unreadable] Relevance: Protein kinases key controllers of cell function, are one of the most important gene families in disease and development of new drug therapies. By exploring how human kinases vary from those of other vertebrates, we can predict whether any human sequence change can predispose to disease or alter drug response, and can distinguish cancer-driving mutations in tumors from background mutations. [unreadable] [unreadable] [unreadable]